zdoom/qzdoom
0
0
Fork 0
mirror of https://github.com/ZDoom/qzdoom.git synced 2024-12-14 06:21:19 +00:00
Code Issues Projects Releases Packages Wiki Activity
qzdoom/src/scripting/vm/vm.h

631 lines
25 KiB
C
Raw Normal View History

- changed Doom source license to GPL in all subdirectories and fixed missing licenses in a few other files.
2017-04-17 10:27:19 +00:00
/*
** vm.h
** VM <-> native interface
**
**---------------------------------------------------------------------------
** Copyright -2016 Randy Heit
** Copyright 2016 Christoph Oelckers
** All rights reserved.
**
** Redistribution and use in source and binary forms, with or without
** modification, are permitted provided that the following conditions
** are met:
**
** 1. Redistributions of source code must retain the above copyright
** notice, this list of conditions and the following disclaimer.
** 2. Redistributions in binary form must reproduce the above copyright
** notice, this list of conditions and the following disclaimer in the
** documentation and/or other materials provided with the distribution.
** 3. The name of the author may not be used to endorse or promote products
** derived from this software without specific prior written permission.
**
** THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
** IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
** OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
** IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
** INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
** NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
** DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
** THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
** (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
** THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**---------------------------------------------------------------------------
**
*/
Normalize line endings
2016-03-01 15:47:10 +00:00
#ifndef VM_H
#define VM_H
#include "zstring.h"
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#include "autosegs.h"
- scriptified a_archvile.cpp. - fixed the type checks for the conditional operator.
2016-10-30 17:41:39 +00:00
#include "vectors.h"
- exported native fields of several more classes.
2016-11-23 00:31:48 +00:00
#include "cmdlib.h"
- record all line numbers during function generation. This is useful for error reporting and eventually debugging. - throw a useful exception when a VM abort occurs, the simple enum was incapable of reporting anything more than the barest minimum, which at least for array index out of bounds errors was insufficient. The current exception mechanism is still insufficient. It really has to report a proper crash location and print a stack trace to the maximum extent possible. Instead it just prints a message and happily goes on. This is not a good solution.
2016-12-02 16:36:29 +00:00
#include "doomerrors.h"
- took VMFunction out of the DObject hierarchy. As it stood, just compiling the internal ZScript code created more than 9000 DObjects, none of which really need to be subjected to garbage collection, aside from allowing lazy deallocation. This puts an incredible drag on the garbage collector which often needs several minutes to finish processing before actual deletion can start. The VM functions with roughly 1800 of these objects were by far the easiest to refactor so they are now. They also use a memory arena now which significantly reduces their memory footprint.
2017-02-08 10:13:41 +00:00
#include "memarena.h"
Did some cleanup - since FScopeBarrier is not in codegen now, it's possible to include it properly from vm.h
2017-03-05 14:05:05 +00:00
#include "scripting/backend/scopebarrier.h"
Normalize line endings
2016-03-01 15:47:10 +00:00
- almost done with TextEnterMenu.
2017-02-18 20:18:23 +00:00
class DObject;
Dynamic virtualscope checking. May yet be buggy.
2017-02-18 04:27:28 +00:00
- Turned DropItem into a plain struct again like it was before the scripting branch got merged. Making this an object had little to no advantage, except being able to remove the deleter code. Now, with some of the class data already being allocated in a memory arena so that freeing it is easier, this can also be used for the drop item lists which makes it unnecessary to subject them to the GC. This also merges the memory arenas for VM functions and flat pointers because both get deleted at the same time so they can share the same one.
2017-02-08 19:37:22 +00:00
extern FMemArena ClassDataAllocator;
Normalize line endings
2016-03-01 15:47:10 +00:00
#define MAX_RETURNS 8 // Maximum number of results a function called by script code can return
#define MAX_TRY_DEPTH 8 // Maximum number of nested TRYs in a single function
typedef unsigned char VM_UBYTE;
typedef signed char VM_SBYTE;
typedef unsigned short VM_UHALF;
typedef signed short VM_SHALF;
typedef unsigned int VM_UWORD;
typedef signed int VM_SWORD;
- changed VM_EPSILON from 1/1024 to 1/65536 so that the ~== operator can actually be used. 1/1024 is far too coarse for ZDoom's purposes, this needs to be at least fixed point precision.
2016-10-22 00:52:28 +00:00
#define VM_EPSILON (1/65536.0)
Normalize line endings
2016-03-01 15:47:10 +00:00
// Register types for VMParam
enum
{
REGT_INT = 0,
REGT_FLOAT = 1,
REGT_STRING = 2,
REGT_POINTER = 3,
REGT_TYPE = 3,
REGT_KONST = 4,
- implemented passing vectors as parameters. So far working for native functions. - removed the bogus optional value from the first A_Jump argument. A quick test with an older ZDoom revealed that this was never working - and implementing it would make things a lot more complicated, especially error checking in the code generator. - fixed: The check for insufficient parameters to a function call was missing.
2016-10-29 11:10:27 +00:00
REGT_MULTIREG2 = 8,
REGT_MULTIREG3 = 16, // (e.g. a vector)
REGT_MULTIREG = 24,
Normalize line endings
2016-03-01 15:47:10 +00:00
REGT_ADDROF = 32, // used with PARAM: pass address of this register
- implemented local vector variables. Currently only the definition plus initial assignment works. - removed all vector4 handling that had already been added, now that this type can no longer be defined.
2016-10-28 13:15:30 +00:00
REGT_NIL = 128 // parameter was omitted
Normalize line endings
2016-03-01 15:47:10 +00:00
};
#define RET_FINAL (0x80) // Used with RET and RETI in the destination slot: this is the final return value
The VM now properly aborts on critical errors
2016-09-14 10:30:11 +00:00
enum EVMAbortException
{
- exported DrawTexture to scripting. This uses templates to avoid source duplication of ParseDrawTextureTags. Not tested yet.
2017-02-04 21:09:49 +00:00
X_OTHER,
The VM now properly aborts on critical errors
2016-09-14 10:30:11 +00:00
X_READ_NIL,
X_WRITE_NIL,
X_TOO_MANY_TRIES,
X_ARRAY_OUT_OF_BOUNDS,
Added "division by zero" to the VM-aborting errors
2016-09-14 10:43:05 +00:00
X_DIVISION_BY_ZERO,
The VM now aborts when a wrong self pointer is used with user variables to avoid random crashes
2016-09-14 11:56:58 +00:00
X_BAD_SELF,
Implemented format() builtin call
2017-01-13 19:44:34 +00:00
X_FORMAT_ERROR
The VM now properly aborts on critical errors
2016-09-14 10:30:11 +00:00
};
- record all line numbers during function generation. This is useful for error reporting and eventually debugging. - throw a useful exception when a VM abort occurs, the simple enum was incapable of reporting anything more than the barest minimum, which at least for array index out of bounds errors was insufficient. The current exception mechanism is still insufficient. It really has to report a proper crash location and print a stack trace to the maximum extent possible. Instead it just prints a message and happily goes on. This is not a good solution.
2016-12-02 16:36:29 +00:00
class CVMAbortException : public CDoomError
{
public:
static FString stacktrace;
redid the exception mechanism for script-side access violations to be of more use for diagnosing problems. The original implementation just printed a mostly information-free message and then went on as if nothing has happened, making it ridiculously easy to write broken code and release it. Changed it to: * Any VMAbortException will now terminate the game session and go back to the console. * It will also print a VM stack trace with all open functions, including source file and line numbers pointing to the problem spots. For this the relevant information had to be added to the VMScriptFunction class. An interesting effect here was that just throwing the exception object increased the VM's Exec function's stack size from 900 bytes to 70kb, because the compiler allocates a separate local buffer for every single instance of the exception object. The obvious solution was to put this part into a subfunction so that it won't pollute the Exec function's own stack frame. Interesting side effect of this: Exec's stack requirement went down from 900 bytes to 600 bytes. This is still on the high side but already a lot better.
2016-12-03 11:23:13 +00:00
CVMAbortException(EVMAbortException reason, const char *moreinfo, va_list ap);
void MaybePrintMessage();
- record all line numbers during function generation. This is useful for error reporting and eventually debugging. - throw a useful exception when a VM abort occurs, the simple enum was incapable of reporting anything more than the barest minimum, which at least for array index out of bounds errors was insufficient. The current exception mechanism is still insufficient. It really has to report a proper crash location and print a stack trace to the maximum extent possible. Instead it just prints a message and happily goes on. This is not a good solution.
2016-12-02 16:36:29 +00:00
};
- exported line_t's functions to the VM.
2017-01-08 14:45:37 +00:00
// This must be a separate function because the VC compiler would otherwise allocate memory on the stack for every separate instance of the exception object that may get thrown.
void ThrowAbortException(EVMAbortException reason, const char *moreinfo, ...);
Normalize line endings
2016-03-01 15:47:10 +00:00
struct VMReturn
{
void *Location;
VM_UBYTE RegType; // Same as VMParam RegType, except REGT_KONST is invalid; only used by asserts
void SetInt(int val)
{
assert(RegType == REGT_INT);
*(int *)Location = val;
}
void SetFloat(double val)
{
assert(RegType == REGT_FLOAT);
*(double *)Location = val;
}
void SetVector(const double val[3])
{
Make sure to use the REGT_MULTIREG flags for opcodes that need it
2016-10-30 05:17:37 +00:00
assert(RegType == (REGT_FLOAT|REGT_MULTIREG3));
Normalize line endings
2016-03-01 15:47:10 +00:00
((double *)Location)[0] = val[0];
((double *)Location)[1] = val[1];
((double *)Location)[2] = val[2];
}
- scriptified a_archvile.cpp. - fixed the type checks for the conditional operator.
2016-10-30 17:41:39 +00:00
void SetVector(const DVector3 &val)
{
assert(RegType == (REGT_FLOAT | REGT_MULTIREG3));
((double *)Location)[0] = val[0];
((double *)Location)[1] = val[1];
((double *)Location)[2] = val[2];
}
- fixed register allocation for vector arguments to script functions. - handle 2D and 3D vectors in SetReturn.
2016-10-29 11:42:37 +00:00
void SetVector2(const double val[2])
{
Make sure to use the REGT_MULTIREG flags for opcodes that need it
2016-10-30 05:17:37 +00:00
assert(RegType == (REGT_FLOAT|REGT_MULTIREG2));
- fixed register allocation for vector arguments to script functions. - handle 2D and 3D vectors in SetReturn.
2016-10-29 11:42:37 +00:00
((double *)Location)[0] = val[0];
((double *)Location)[1] = val[1];
}
- scriptified a_archvile.cpp. - fixed the type checks for the conditional operator.
2016-10-30 17:41:39 +00:00
void SetVector2(const DVector2 &val)
{
assert(RegType == (REGT_FLOAT | REGT_MULTIREG2));
((double *)Location)[0] = val[0];
((double *)Location)[1] = val[1];
}
Normalize line endings
2016-03-01 15:47:10 +00:00
void SetString(const FString &val)
{
assert(RegType == REGT_STRING);
*(FString *)Location = val;
}
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
void SetPointer(void *val)
{
assert(RegType == REGT_POINTER);
*(void **)Location = val;
}
void SetObject(DObject *val)
{
assert(RegType == REGT_POINTER);
*(void **)Location = val;
}
Normalize line endings
2016-03-01 15:47:10 +00:00
void IntAt(int *loc)
{
Location = loc;
RegType = REGT_INT;
}
void FloatAt(double *loc)
{
Location = loc;
RegType = REGT_FLOAT;
}
void StringAt(FString *loc)
{
Location = loc;
RegType = REGT_STRING;
}
void PointerAt(void **loc)
{
Location = loc;
RegType = REGT_POINTER;
}
- eliminated the native PowerupGiver class. - scriptified the respawn invulnerability code into a virtual OnRespawn function for PlayerPawn so that custom effects can be implemented.
2017-01-17 23:11:04 +00:00
VMReturn() { }
VMReturn(int *loc) { IntAt(loc); }
VMReturn(double *loc) { FloatAt(loc); }
VMReturn(FString *loc) { StringAt(loc); }
VMReturn(void **loc) { PointerAt(loc); }
Normalize line endings
2016-03-01 15:47:10 +00:00
};
struct VMRegisters;
struct VMValue
{
union
{
int i;
- removed the last remnants of the ATAGs.
2017-04-10 15:08:52 +00:00
void *a;
Normalize line endings
2016-03-01 15:47:10 +00:00
double f;
struct { int pad[3]; VM_UBYTE Type; };
struct { int foo[4]; } biggest;
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
const FString *sp;
Normalize line endings
2016-03-01 15:47:10 +00:00
};
// Unfortunately, FString cannot be used directly.
// Fortunately, it is relatively simple.
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
const FString &s() const { return *sp; }
Normalize line endings
2016-03-01 15:47:10 +00:00
VMValue()
{
a = NULL;
Type = REGT_NIL;
}
VMValue(const VMValue &o)
{
biggest = o.biggest;
}
VMValue(int v)
{
i = v;
Type = REGT_INT;
}
VMValue(double v)
{
f = v;
Type = REGT_FLOAT;
}
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
VMValue(const char *s) = delete;
VMValue(const FString &s) = delete;
VMValue(const FString *s)
Normalize line endings
2016-03-01 15:47:10 +00:00
{
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
sp = s;
Normalize line endings
2016-03-01 15:47:10 +00:00
Type = REGT_STRING;
}
VMValue(DObject *v)
{
a = v;
Type = REGT_POINTER;
}
VMValue(void *v)
{
a = v;
Type = REGT_POINTER;
}
Restored explicit assignment operators for VMValue Unfortunately strings require special handling and so all operators must be available This partly solves https://mantis.zdoom.org/view.php?id=453 (and https://mantis.zdoom.org/view.php?id=459 too)
2017-03-21 09:46:33 +00:00
VMValue &operator=(const VMValue &o)
{
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
biggest = o.biggest;
Restored explicit assignment operators for VMValue Unfortunately strings require special handling and so all operators must be available This partly solves https://mantis.zdoom.org/view.php?id=453 (and https://mantis.zdoom.org/view.php?id=459 too)
2017-03-21 09:46:33 +00:00
return *this;
}
VMValue &operator=(int v)
{
i = v;
Type = REGT_INT;
return *this;
}
VMValue &operator=(double v)
{
f = v;
Type = REGT_FLOAT;
return *this;
}
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
VMValue &operator=(const FString *v)
Restored explicit assignment operators for VMValue Unfortunately strings require special handling and so all operators must be available This partly solves https://mantis.zdoom.org/view.php?id=453 (and https://mantis.zdoom.org/view.php?id=459 too)
2017-03-21 09:46:33 +00:00
{
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
sp = v;
Type = REGT_STRING;
Restored explicit assignment operators for VMValue Unfortunately strings require special handling and so all operators must be available This partly solves https://mantis.zdoom.org/view.php?id=453 (and https://mantis.zdoom.org/view.php?id=459 too)
2017-03-21 09:46:33 +00:00
return *this;
}
- changed VMValue to handle strings by reference. This makes VMValue a real POD type with no hacky overloads and eliminates a lot of destructor code in all places that call a VM function. Due to the way this had to be handled, none of these destructors could be skipped because any value could have been a string. This required some minor changes in functions that passed a temporary FString into the VM to ensure that the temporary object lives long enough to be handled. The code generator had already been changed to deal with this in a previous commit. This is easily offset by the code savings and reduced maintenance needs elsewhere.
2017-03-22 00:44:56 +00:00
VMValue &operator=(const FString &v) = delete;
VMValue &operator=(const char *v) = delete;
Restored explicit assignment operators for VMValue Unfortunately strings require special handling and so all operators must be available This partly solves https://mantis.zdoom.org/view.php?id=453 (and https://mantis.zdoom.org/view.php?id=459 too)
2017-03-21 09:46:33 +00:00
VMValue &operator=(DObject *v)
{
a = v;
Type = REGT_POINTER;
return *this;
}
Normalize line endings
2016-03-01 15:47:10 +00:00
int ToInt()
{
if (Type == REGT_INT)
{
return i;
}
if (Type == REGT_FLOAT)
{
return int(f);
}
if (Type == REGT_STRING)
{
return s().ToLong();
}
// FIXME
return 0;
}
double ToDouble()
{
if (Type == REGT_FLOAT)
{
return f;
}
if (Type == REGT_INT)
{
return i;
}
if (Type == REGT_STRING)
{
return s().ToDouble();
}
// FIXME
return 0;
}
};
- took VMFunction out of the DObject hierarchy. As it stood, just compiling the internal ZScript code created more than 9000 DObjects, none of which really need to be subjected to garbage collection, aside from allowing lazy deallocation. This puts an incredible drag on the garbage collector which often needs several minutes to finish processing before actual deletion can start. The VM functions with roughly 1800 of these objects were by far the easiest to refactor so they are now. They also use a memory arena now which significantly reduces their memory footprint.
2017-02-08 10:13:41 +00:00
class VMFunction
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
{
public:
Implemented implicit 'protected' inheritance in virtual functions
2017-03-04 23:24:22 +00:00
bool Unsafe = false;
int VarFlags = 0; // [ZZ] this replaces 5+ bool fields
- replaced another large batch of homegrown type use.
2017-03-08 17:44:37 +00:00
uint8_t ImplicitArgs = 0; // either 0 for static, 1 for method or 3 for action
- disabled the scripted virtual function module after finding out that it only works if each single class that may serve as a parent for scripting is explicitly declared. Needless to say, this is simply too volatile and would require constant active maintenance, not to mention a huge amount of work up front to get going. It also hid a nasty problem with the Destroy method. Due to the way the garbage collector works, Destroy cannot be exposed to scripts as-is. It may be called from scripts but it may not be overridden from scripts because the garbage collector can call this function after all data needed for calling a scripted override has already been destroyed because if that data is also being collected there is no guarantee that proper order of destruction is observed. So for now Destroy is just a normal native method to scripted classes
2016-11-24 20:36:02 +00:00
unsigned VirtualIndex = ~0u;
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
FName Name;
- fixed: A VM function's NumArgs value needs to count stack arguments, not logical ones, meaning that for vectors each element needs to count separately. - renamed VMFunction::Defaults to DefaultArgs to make searching easier. - let ZCCCompiler process vector defaults for function parameters.
2016-10-30 08:05:42 +00:00
TArray<VMValue> DefaultArgs;
- scriptified the scripted marines. - fixed symbol name generation for native functions. - moved PrintableName to VMFunction so that native functions also have this information.
2016-11-21 18:09:58 +00:00
FString PrintableName; // so that the VM can print meaningful info if something in this function goes wrong.
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
class PPrototype *Proto;
Implemented implicit 'protected' inheritance in virtual functions
2017-03-04 23:24:22 +00:00
VMFunction(FName name = NAME_None) : ImplicitArgs(0), Name(name), Proto(NULL)
- took VMFunction out of the DObject hierarchy. As it stood, just compiling the internal ZScript code created more than 9000 DObjects, none of which really need to be subjected to garbage collection, aside from allowing lazy deallocation. This puts an incredible drag on the garbage collector which often needs several minutes to finish processing before actual deletion can start. The VM functions with roughly 1800 of these objects were by far the easiest to refactor so they are now. They also use a memory arena now which significantly reduces their memory footprint.
2017-02-08 10:13:41 +00:00
{
AllFunctions.Push(this);
}
virtual ~VMFunction() {}
void *operator new(size_t size)
{
- Turned DropItem into a plain struct again like it was before the scripting branch got merged. Making this an object had little to no advantage, except being able to remove the deleter code. Now, with some of the class data already being allocated in a memory arena so that freeing it is easier, this can also be used for the drop item lists which makes it unnecessary to subject them to the GC. This also merges the memory arenas for VM functions and flat pointers because both get deleted at the same time so they can share the same one.
2017-02-08 19:37:22 +00:00
return ClassDataAllocator.Alloc(size);
- took VMFunction out of the DObject hierarchy. As it stood, just compiling the internal ZScript code created more than 9000 DObjects, none of which really need to be subjected to garbage collection, aside from allowing lazy deallocation. This puts an incredible drag on the garbage collector which often needs several minutes to finish processing before actual deletion can start. The VM functions with roughly 1800 of these objects were by far the easiest to refactor so they are now. They also use a memory arena now which significantly reduces their memory footprint.
2017-02-08 10:13:41 +00:00
}
void operator delete(void *block) {}
void operator delete[](void *block) {}
static void DeleteAll()
{
for (auto f : AllFunctions)
{
f->~VMFunction();
}
AllFunctions.Clear();
}
static TArray<VMFunction *> AllFunctions;
protected:
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
};
Normalize line endings
2016-03-01 15:47:10 +00:00
class VMNativeFunction : public VMFunction
{
public:
- allow the VM to run on one global stack per thread. It is utterly pointless to require every function that wants to make a VM call to allocate a new stack first. The allocation overhead doubles the time to set up the call. With one stack, previously allocated memory can be reused. The only important thing is, if this ever gets used in a multithreaded environment to have the stack being declared as thread_local, although for ZDoom this is of no consequence. - eliminated all cases where native code was calling other native code through the VM interface. After scriptifying the game code, only 5 places were left which were quickly eliminated. This was mostly to ensure that the native VM function parameters do not need to be propagated further than absolutely necessary.
2016-11-30 16:15:01 +00:00
typedef int (*NativeCallType)(VMValue *param, TArray<VMValue> &defaultparam, int numparam, VMReturn *ret, int numret);
Normalize line endings
2016-03-01 15:47:10 +00:00
Removed BarrierSide from VMFunction, since it's obsolete with addition of VarFlags
2017-03-04 23:31:23 +00:00
// 8 is VARF_Native. I can't write VARF_Native because of circular references between this and dobject/dobjtype.
Implemented implicit 'protected' inheritance in virtual functions
2017-03-04 23:24:22 +00:00
VMNativeFunction() : NativeCall(NULL) { VarFlags = 8; }
VMNativeFunction(NativeCallType call) : NativeCall(call) { VarFlags = 8; }
VMNativeFunction(NativeCallType call, FName name) : VMFunction(name), NativeCall(call) { VarFlags = 8; }
Normalize line endings
2016-03-01 15:47:10 +00:00
// Return value is the number of results.
NativeCallType NativeCall;
};
- moved the VM types into their own file and only include it where really needed.
2017-04-12 23:12:04 +00:00
int VMCall(VMFunction *func, VMValue *params, int numparams, VMReturn *results, int numresults/*, VMException **trap = NULL*/);
Normalize line endings
2016-03-01 15:47:10 +00:00
// Use this in the prototype for a native function.
- allow the VM to run on one global stack per thread. It is utterly pointless to require every function that wants to make a VM call to allocate a new stack first. The allocation overhead doubles the time to set up the call. With one stack, previously allocated memory can be reused. The only important thing is, if this ever gets used in a multithreaded environment to have the stack being declared as thread_local, although for ZDoom this is of no consequence. - eliminated all cases where native code was calling other native code through the VM interface. After scriptifying the game code, only 5 places were left which were quickly eliminated. This was mostly to ensure that the native VM function parameters do not need to be propagated further than absolutely necessary.
2016-11-30 16:15:01 +00:00
#define VM_ARGS VMValue *param, TArray<VMValue> &defaultparam, int numparam, VMReturn *ret, int numret
#define VM_ARGS_NAMES param, defaultparam, numparam, ret, numret
Normalize line endings
2016-03-01 15:47:10 +00:00
// Use these to collect the parameters in a native function.
// variable name <x> at position <p>
- added null pointer validation to any relevant exported function. In most cases null pointers were already being treated as 'do nothing', but there's several places where this can make the code silently fail so in these cases a VM exception will be raised, once the VM's exception handling has been repaired to provide useful diagnostics. (Right now all it does is catch the exception, print a useless message and return to the caller as if nothing has happened.)
2016-12-02 11:06:49 +00:00
void NullParam(const char *varname);
Made usage of assertions in VM code consistent https://forum.zdoom.org/viewtopic.php?t=56995
2017-06-22 06:45:09 +00:00
#ifndef NDEBUG
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
bool AssertObject(void * ob);
#endif
- added null pointer validation to any relevant exported function. In most cases null pointers were already being treated as 'do nothing', but there's several places where this can make the code silently fail so in these cases a VM exception will be raised, once the VM's exception handling has been repaired to provide useful diagnostics. (Right now all it does is catch the exception, print a useless message and return to the caller as if nothing has happened.)
2016-12-02 11:06:49 +00:00
#define PARAM_NULLCHECK(ptr, var) (ptr == nullptr? NullParam(#var), ptr : ptr)
Normalize line endings
2016-03-01 15:47:10 +00:00
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
#define ASSERTINT(p) assert((p).Type == REGT_INT)
#define ASSERTFLOAT(p) assert((p).Type == REGT_FLOAT)
#define ASSERTSTRING(p) assert((p).Type == REGT_STRING)
#define ASSERTOBJECT(p) assert((p).Type == REGT_POINTER && AssertObject(p.a))
#define ASSERTPOINTER(p) assert((p).Type == REGT_POINTER)
Normalize line endings
2016-03-01 15:47:10 +00:00
// For required parameters.
#define PARAM_INT_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); int x = param[p].i;
- implemented the backend for dynamic arrays. Still needs thorough testing but it should be complete. - use a memory arena to store flat pointers so that the messed up cleanup can be avoided by deallocating this in bulk. - added a new SO opcode to the VM to execute a write barrier. This is necessary for all objects that are not linked into one global table, i.e. everything except thinkers and class types. - always use the cheaper LOS opcode for reading pointers to classes and defaults because these cannot be destroyed during normal operation. - removed the pointless validation from String.Mid. If the values are read as unsigned the internal validation of FString::Mid will automatically ensure proper results.
2017-02-07 13:48:27 +00:00
#define PARAM_UINT_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); unsigned x = param[p].i;
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_BOOL_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); bool x = !!param[p].i;
#define PARAM_NAME_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); FName x = ENamedName(param[p].i);
#define PARAM_SOUND_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); FSoundID x = param[p].i;
#define PARAM_COLOR_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); PalEntry x; x.d = param[p].i;
#define PARAM_FLOAT_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_FLOAT); double x = param[p].f;
- added string handling to comparison operators.
2016-10-29 16:57:56 +00:00
#define PARAM_ANGLE_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_FLOAT); DAngle x = param[p].f;
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_STRING_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_STRING); FString x = param[p].s();
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
#define PARAM_STATE_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); FState *x = (FState *)StateLabels.GetState(param[p].i, self->GetClass());
#define PARAM_STATE_ACTION_AT(p,x) assert((p) < numparam); assert(param[p].Type == REGT_INT); FState *x = (FState *)StateLabels.GetState(param[p].i, stateowner->GetClass());
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_POINTER_AT(p,x,type) assert((p) < numparam); assert(param[p].Type == REGT_POINTER); type *x = (type *)param[p].a;
- scriptified the input functions of DMessageBoxMenu.
2017-02-18 22:05:01 +00:00
#define PARAM_POINTERTYPE_AT(p,x,type) assert((p) < numparam); assert(param[p].Type == REGT_POINTER); type x = (type )param[p].a;
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
#define PARAM_OBJECT_AT(p,x,type) assert((p) < numparam); assert(param[p].Type == REGT_POINTER && AssertObject(param[p].a)); type *x = (type *)param[p].a; assert(x == NULL || x->IsKindOf(RUNTIME_CLASS(type)));
- separated class descriptors from VM types. Combining these two groups of data has been the cause of many hard to detect errors because it allowed liberal casting between types that are used for completely different things.
2017-04-12 20:46:49 +00:00
#define PARAM_CLASS_AT(p,x,base) assert((p) < numparam); assert(param[p].Type == REGT_POINTER); base::MetaClass *x = (base::MetaClass *)param[p].a; assert(x == NULL || x->IsDescendantOf(RUNTIME_CLASS(base)));
- added null pointer validation to any relevant exported function. In most cases null pointers were already being treated as 'do nothing', but there's several places where this can make the code silently fail so in these cases a VM exception will be raised, once the VM's exception handling has been repaired to provide useful diagnostics. (Right now all it does is catch the exception, print a useless message and return to the caller as if nothing has happened.)
2016-12-02 11:06:49 +00:00
#define PARAM_POINTER_NOT_NULL_AT(p,x,type) assert((p) < numparam); assert(param[p].Type == REGT_POINTER); type *x = (type *)PARAM_NULLCHECK(param[p].a, #x);
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
#define PARAM_OBJECT_NOT_NULL_AT(p,x,type) assert((p) < numparam); assert(param[p].Type == REGT_POINTER && (AssertObject(param[p].a))); type *x = (type *)PARAM_NULLCHECK(param[p].a, #x); assert(x == NULL || x->IsKindOf(RUNTIME_CLASS(type)));
- separated class descriptors from VM types. Combining these two groups of data has been the cause of many hard to detect errors because it allowed liberal casting between types that are used for completely different things.
2017-04-12 20:46:49 +00:00
#define PARAM_CLASS_NOT_NULL_AT(p,x,base) assert((p) < numparam); assert(param[p].Type == REGT_POINTER); base::MetaClass *x = (base::MetaClass *)PARAM_NULLCHECK(param[p].a, #x); assert(x == NULL || x->IsDescendantOf(RUNTIME_CLASS(base)));
Normalize line endings
2016-03-01 15:47:10 +00:00
- removed the REGT_NIL check from the PARAM_EXISTS macro, because with the implemented handling of named arguments it isn't possible that REGT_NIL can end up in the parameters without a programming error..
2016-11-21 09:28:06 +00:00
#define PARAM_EXISTS(p) ((p) < numparam)
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
#define PARAM_INT_DEF_AT(p,x) int x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = param[p].i; } else { ASSERTINT(defaultparam[p]); x = defaultparam[p].i; }
#define PARAM_BOOL_DEF_AT(p,x) bool x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = !!param[p].i; } else { ASSERTINT(defaultparam[p]); x = !!defaultparam[p].i; }
#define PARAM_NAME_DEF_AT(p,x) FName x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = ENamedName(param[p].i); } else { ASSERTINT(defaultparam[p]); x = ENamedName(defaultparam[p].i); }
#define PARAM_SOUND_DEF_AT(p,x) FSoundID x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = FSoundID(param[p].i); } else { ASSERTINT(defaultparam[p]); x = FSoundID(defaultparam[p].i); }
#define PARAM_COLOR_DEF_AT(p,x) PalEntry x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = param[p].i; } else { ASSERTINT(defaultparam[p]); x = defaultparam[p].i; }
#define PARAM_FLOAT_DEF_AT(p,x) double x; if (PARAM_EXISTS(p)) { ASSERTFLOAT(param[p]); x = param[p].f; } else { ASSERTFLOAT(defaultparam[p]); x = defaultparam[p].f; }
#define PARAM_ANGLE_DEF_AT(p,x) DAngle x; if (PARAM_EXISTS(p)) { ASSERTFLOAT(param[p]); x = param[p].f; } else { ASSERTFLOAT(defaultparam[p]); x = defaultparam[p].f; }
- scriptified the main statusbar interface and the Strife status bar. Note that the Strife status bar does not draw the health bars yet. I tried to replace the hacky custom texture with a single fill operation but had to find out that all the coordinate mangling for the status bar is being done deep in the video code. This needs to be fixed before this can be made to work. Currently this is not usable in mods because they cannot initialize custom status bars yet.
2017-03-21 19:09:09 +00:00
#define PARAM_STRING_DEF_AT(p,x) FString x; if (PARAM_EXISTS(p)) { ASSERTSTRING(param[p]); x = param[p].s(); } else { ASSERTSTRING(defaultparam[p]); x = defaultparam[p].s(); }
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
#define PARAM_STATE_DEF_AT(p,x) FState *x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = (FState*)StateLabels.GetState(param[p].i, self->GetClass()); } else { ASSERTINT(defaultparam[p]); x = (FState*)StateLabels.GetState(defaultparam[p].i, self->GetClass()); }
#define PARAM_STATE_ACTION_DEF_AT(p,x) FState *x; if (PARAM_EXISTS(p)) { ASSERTINT(param[p]); x = (FState*)StateLabels.GetState(param[p].i, stateowner->GetClass()); } else { ASSERTINT(defaultparam[p]); x = (FState*)StateLabels.GetState(defaultparam[p].i, stateowner->GetClass()); }
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
#define PARAM_POINTER_DEF_AT(p,x,t) t *x; if (PARAM_EXISTS(p)) { ASSERTPOINTER(param[p]); x = (t*)param[p].a; } else { ASSERTPOINTER(defaultparam[p]); x = (t*)defaultparam[p].a; }
#define PARAM_OBJECT_DEF_AT(p,x,t) t *x; if (PARAM_EXISTS(p)) { ASSERTOBJECT(param[p]); x = (t*)param[p].a; } else { ASSERTOBJECT(defaultparam[p]); x = (t*)defaultparam[p].a; }
- separated class descriptors from VM types. Combining these two groups of data has been the cause of many hard to detect errors because it allowed liberal casting between types that are used for completely different things.
2017-04-12 20:46:49 +00:00
#define PARAM_CLASS_DEF_AT(p,x,t) t::MetaClass *x; if (PARAM_EXISTS(p)) { x = (t::MetaClass*)param[p].a; } else { ASSERTPOINTER(defaultparam[p]); x = (t::MetaClass*)defaultparam[p].a; }
#define PARAM_CLASS_DEF_NOT_NULL_AT(p,x,t) t::MetaClass *x; if (PARAM_EXISTS(p)) { x = (t::MetaClass*)PARAM_NULLCHECK(param[p].a, #x); } else { x = (t::MetaClass*)PARAM_NULLCHECK(defaultparam[p].a, #x); }
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
Normalize line endings
2016-03-01 15:47:10 +00:00
// The above, but with an automatically increasing position index.
#define PARAM_PROLOGUE int paramnum = -1;
#define PARAM_INT(x) ++paramnum; PARAM_INT_AT(paramnum,x)
- implemented the backend for dynamic arrays. Still needs thorough testing but it should be complete. - use a memory arena to store flat pointers so that the messed up cleanup can be avoided by deallocating this in bulk. - added a new SO opcode to the VM to execute a write barrier. This is necessary for all objects that are not linked into one global table, i.e. everything except thinkers and class types. - always use the cheaper LOS opcode for reading pointers to classes and defaults because these cannot be destroyed during normal operation. - removed the pointless validation from String.Mid. If the values are read as unsigned the internal validation of FString::Mid will automatically ensure proper results.
2017-02-07 13:48:27 +00:00
#define PARAM_UINT(x) ++paramnum; PARAM_UINT_AT(paramnum,x)
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_BOOL(x) ++paramnum; PARAM_BOOL_AT(paramnum,x)
#define PARAM_NAME(x) ++paramnum; PARAM_NAME_AT(paramnum,x)
#define PARAM_SOUND(x) ++paramnum; PARAM_SOUND_AT(paramnum,x)
#define PARAM_COLOR(x) ++paramnum; PARAM_COLOR_AT(paramnum,x)
#define PARAM_FLOAT(x) ++paramnum; PARAM_FLOAT_AT(paramnum,x)
- renamed PARAM_DANGLE macros to PARAM_ANGLE, now that the old angle_t type is no longer used in the VM interface.
2016-03-25 12:23:07 +00:00
#define PARAM_ANGLE(x) ++paramnum; PARAM_ANGLE_AT(paramnum,x)
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_STRING(x) ++paramnum; PARAM_STRING_AT(paramnum,x)
#define PARAM_STATE(x) ++paramnum; PARAM_STATE_AT(paramnum,x)
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
#define PARAM_STATE_ACTION(x) ++paramnum; PARAM_STATE_ACTION_AT(paramnum,x)
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_POINTER(x,type) ++paramnum; PARAM_POINTER_AT(paramnum,x,type)
- scriptified the input functions of DMessageBoxMenu.
2017-02-18 22:05:01 +00:00
#define PARAM_POINTERTYPE(x,type) ++paramnum; PARAM_POINTERTYPE_AT(paramnum,x,type)
Normalize line endings
2016-03-01 15:47:10 +00:00
#define PARAM_OBJECT(x,type) ++paramnum; PARAM_OBJECT_AT(paramnum,x,type)
#define PARAM_CLASS(x,base) ++paramnum; PARAM_CLASS_AT(paramnum,x,base)
- some conceptual work on statusbar stuff. This looks like a viable approach to build something that can replace SBARINFO.
2017-03-22 18:56:21 +00:00
#define PARAM_CLASS(x,base) ++paramnum; PARAM_CLASS_AT(paramnum,x,base)
- added null pointer validation to any relevant exported function. In most cases null pointers were already being treated as 'do nothing', but there's several places where this can make the code silently fail so in these cases a VM exception will be raised, once the VM's exception handling has been repaired to provide useful diagnostics. (Right now all it does is catch the exception, print a useless message and return to the caller as if nothing has happened.)
2016-12-02 11:06:49 +00:00
#define PARAM_POINTER_NOT_NULL(x,type) ++paramnum; PARAM_POINTER_NOT_NULL_AT(paramnum,x,type)
#define PARAM_OBJECT_NOT_NULL(x,type) ++paramnum; PARAM_OBJECT_NOT_NULL_AT(paramnum,x,type)
#define PARAM_CLASS_NOT_NULL(x,base) ++paramnum; PARAM_CLASS_NOT_NULL_AT(paramnum,x,base)
Normalize line endings
2016-03-01 15:47:10 +00:00
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
#define PARAM_INT_DEF(x) ++paramnum; PARAM_INT_DEF_AT(paramnum,x)
#define PARAM_BOOL_DEF(x) ++paramnum; PARAM_BOOL_DEF_AT(paramnum,x)
#define PARAM_NAME_DEF(x) ++paramnum; PARAM_NAME_DEF_AT(paramnum,x)
#define PARAM_SOUND_DEF(x) ++paramnum; PARAM_SOUND_DEF_AT(paramnum,x)
#define PARAM_COLOR_DEF(x) ++paramnum; PARAM_COLOR_DEF_AT(paramnum,x)
#define PARAM_FLOAT_DEF(x) ++paramnum; PARAM_FLOAT_DEF_AT(paramnum,x)
#define PARAM_ANGLE_DEF(x) ++paramnum; PARAM_ANGLE_DEF_AT(paramnum,x)
#define PARAM_STRING_DEF(x) ++paramnum; PARAM_STRING_DEF_AT(paramnum,x)
#define PARAM_STATE_DEF(x) ++paramnum; PARAM_STATE_DEF_AT(paramnum,x)
- fixed: State labels were resolved in the calling function's context instead of the called function one's. This could cause problems with functions that take states as parameters but use them to set them internally instead of passing them through the A_Jump interface back to the caller, like A_Chase or A_LookEx. This required some quite significant refactoring because the entire state resolution logic had been baked into the compiler which turned out to be a major maintenance problem. Fixed this by adding a new builtin type 'statelabel'. This is an opaque identifier representing a state, with the actual data either directly encoded into the number for single label state or an index into a state information table. The state resolution is now the task of the called function as it should always have remained. Note, that this required giving back the 'action' qualifier to most state jumping functions. - refactored most A_Jump checkers to a two stage setup with a pure checker that returns a boolean and a scripted A_Jump wrapper, for some simpler checks the checker function was entirely omitted and calculated inline in the A_Jump function. It is strongly recommended to use the boolean checkers unless using an inline function invocation in a state as they lead to vastly clearer code and offer more flexibility. - let Min() and Max() use the OP_MIN and OP_MAX opcodes. Although these were present, these function were implemented using some grossly inefficient branching tests. - the DECORATE 'state' cast kludge will now actually call ResolveState because a state label is not a state and needs conversion.
2016-11-14 13:12:27 +00:00
#define PARAM_STATE_ACTION_DEF(x) ++paramnum; PARAM_STATE_ACTION_DEF_AT(paramnum,x)
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
#define PARAM_POINTER_DEF(x,type) ++paramnum; PARAM_POINTER_DEF_AT(paramnum,x,type)
#define PARAM_OBJECT_DEF(x,type) ++paramnum; PARAM_OBJECT_DEF_AT(paramnum,x,type)
#define PARAM_CLASS_DEF(x,base) ++paramnum; PARAM_CLASS_DEF_AT(paramnum,x,base)
- added null pointer validation to any relevant exported function. In most cases null pointers were already being treated as 'do nothing', but there's several places where this can make the code silently fail so in these cases a VM exception will be raised, once the VM's exception handling has been repaired to provide useful diagnostics. (Right now all it does is catch the exception, print a useless message and return to the caller as if nothing has happened.)
2016-12-02 11:06:49 +00:00
#define PARAM_CLASS_DEF_NOT_NULL(x,base) ++paramnum; PARAM_CLASS_DEF_NOT_NULL_AT(paramnum,x,base)
- actually evaluate the default parameters and store them in the VMFunction. - disabled the assert in PType::GetRegType. This assert blocks any use to check for types that are incompatible with function parameters. - pass the default parameter constants to the native functions. At the moment this is not used yet. - use the function defaults to complete argument lists to script functions. - fixed all default values that got flagged by the expression evaluator as non-constant. Most were state labels and colors which were defaulted to "". The proper value is null for states and 0 for colors. - also replaced all "" defaults for names with "none".
2016-10-26 23:30:34 +00:00
- allow the VM to run on one global stack per thread. It is utterly pointless to require every function that wants to make a VM call to allocate a new stack first. The allocation overhead doubles the time to set up the call. With one stack, previously allocated memory can be reused. The only important thing is, if this ever gets used in a multithreaded environment to have the stack being declared as thread_local, although for ZDoom this is of no consequence. - eliminated all cases where native code was calling other native code through the VM interface. After scriptifying the game code, only 5 places were left which were quickly eliminated. This was mostly to ensure that the native VM function parameters do not need to be propagated further than absolutely necessary.
2016-11-30 16:15:01 +00:00
typedef int(*actionf_p)(VMValue *param, TArray<VMValue> &defaultparam, int numparam, VMReturn *ret, int numret);/*(VM_ARGS)*/
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
struct FieldDesc
{
const char *ClassName;
const char *FieldName;
- fixed variable size mismatch and some warnings for 32 bit builds.
2017-03-14 17:48:51 +00:00
size_t FieldOffset;
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
unsigned FieldSize;
int BitValue;
};
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
struct AFuncDesc
{
- scriptified the scripted marines. - fixed symbol name generation for native functions. - moved PrintableName to VMFunction so that native functions also have this information.
2016-11-21 18:09:58 +00:00
const char *ClassName;
const char *FuncName;
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
actionf_p Function;
VMNativeFunction **VMPointer;
};
#if defined(_MSC_VER)
#pragma section(".areg$u",read)
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#pragma section(".freg$u",read)
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#define MSVC_ASEG __declspec(allocate(".areg$u"))
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#define MSVC_FSEG __declspec(allocate(".freg$u"))
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#define GCC_ASEG
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#define GCC_FSEG
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#else
#define MSVC_ASEG
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#define MSVC_FSEG
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#define GCC_ASEG __attribute__((section(SECTION_AREG))) __attribute__((used))
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#define GCC_FSEG __attribute__((section(SECTION_FREG))) __attribute__((used))
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#endif
// Macros to handle action functions. These are here so that I don't have to
// change every single use in case the parameters change.
#define DEFINE_ACTION_FUNCTION(cls, name) \
- scriptified two of the Acolyte's functions. - added a DActorIterator class. - fixed: It was not possible to have functions of the same name in two different classes because the name they were searched for was not qualified by the class. Changed so that the class name is included now, but to avoid renaming several hundreds of functions all at once, if the search fails, it will repeat with 'Actor' as class name. This commit contains preparations for scriptifying Hexen's Dragon, but that doesn't work yet so it's not included.
2016-11-16 00:36:21 +00:00
static int AF_##cls##_##name(VM_ARGS); \
VMNativeFunction *cls##_##name##_VMPtr; \
- scriptified the scripted marines. - fixed symbol name generation for native functions. - moved PrintableName to VMFunction so that native functions also have this information.
2016-11-21 18:09:58 +00:00
static const AFuncDesc cls##_##name##_Hook = { #cls, #name, AF_##cls##_##name, &cls##_##name##_VMPtr }; \
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
extern AFuncDesc const *const cls##_##name##_HookPtr; \
MSVC_ASEG AFuncDesc const *const cls##_##name##_HookPtr GCC_ASEG = &cls##_##name##_Hook; \
- scriptified two of the Acolyte's functions. - added a DActorIterator class. - fixed: It was not possible to have functions of the same name in two different classes because the name they were searched for was not qualified by the class. Changed so that the class name is included now, but to avoid renaming several hundreds of functions all at once, if the search fails, it will repeat with 'Actor' as class name. This commit contains preparations for scriptifying Hexen's Dragon, but that doesn't work yet so it's not included.
2016-11-16 00:36:21 +00:00
static int AF_##cls##_##name(VM_ARGS)
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
// cls is the scripted class name, icls the internal one (e.g. player_t vs. Player)
#define DEFINE_FIELD_X(cls, icls, name) \
static const FieldDesc VMField_##icls##_##name = { "A" #cls, #name, (unsigned)myoffsetof(icls, name), (unsigned)sizeof(icls::name), 0 }; \
extern FieldDesc const *const VMField_##icls##_##name##_HookPtr; \
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
MSVC_FSEG FieldDesc const *const VMField_##icls##_##name##_HookPtr GCC_FSEG = &VMField_##icls##_##name;
- fixed variable size mismatch and some warnings for 32 bit builds.
2017-03-14 17:48:51 +00:00
// This is for cases where the internal size does not match the part that gets exported.
#define DEFINE_FIELD_UNSIZED(cls, icls, name) \
static const FieldDesc VMField_##icls##_##name = { "A" #cls, #name, (unsigned)myoffsetof(icls, name), ~0u, 0 }; \
extern FieldDesc const *const VMField_##icls##_##name##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##icls##_##name##_HookPtr GCC_FSEG = &VMField_##icls##_##name;
- added proper definitions for all exported native fields. - synthesize native fields for all declared flags, not just for AActor.
2016-11-22 22:42:32 +00:00
#define DEFINE_FIELD_NAMED_X(cls, icls, name, scriptname) \
- added the ability to attach a constructor or destructor to an internally defined struct. There are a few which require explicit native construction or destruction that need to be exported to the VM, e.g. FCheckPosition. The VM cannot handle this directly, it needs two special functions to be attached to handle such elements.
2017-01-11 22:46:03 +00:00
static const FieldDesc VMField_##cls##_##scriptname = { "A" #cls, #scriptname, (unsigned)myoffsetof(icls, name), (unsigned)sizeof(icls::name), 0 }; \
extern FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr GCC_FSEG = &VMField_##cls##_##scriptname;
- allow defining native fields through scripts. Internally this only requires exporting the address, but not the entire field. - added new VARF_Transient flag so that the decision whether to serialize a field does not depend solely on its native status. It may actually make a lot of sense to use the auto-serializer for native fields, too, as this would eliminate a lot of maintenance code. - defined (u)int8/16 as aliases to the byte and short types (Can't we not just get rid of this naming convention already...?) - exporting the fields of Actor revealed a few name clashes between them and some global types, so Actor.Sector was renamed to CurSector and Actor.Inventory was renamed to Actor.Inv.
2016-11-22 18:20:31 +00:00
#define DEFINE_FIELD_X_BIT(cls, icls, name, bitval) \
static const FieldDesc VMField_##icls##_##name = { "A" #cls, #name, (unsigned)myoffsetof(icls, name), (unsigned)sizeof(icls::name), bitval }; \
extern FieldDesc const *const VMField_##icls##_##name##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##icls##_##name##_HookPtr GCC_FSEG = &VMField_##cls##_##name;
#define DEFINE_FIELD(cls, name) \
static const FieldDesc VMField_##cls##_##name = { #cls, #name, (unsigned)myoffsetof(cls, name), (unsigned)sizeof(cls::name), 0 }; \
extern FieldDesc const *const VMField_##cls##_##name##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##cls##_##name##_HookPtr GCC_FSEG = &VMField_##cls##_##name;
#define DEFINE_FIELD_NAMED(cls, name, scriptname) \
static const FieldDesc VMField_##cls##_##scriptname = { #cls, #scriptname, (unsigned)myoffsetof(cls, name), (unsigned)sizeof(cls::name), 0 }; \
extern FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr GCC_FSEG = &VMField_##cls##_##scriptname;
#define DEFINE_FIELD_BIT(cls, name, scriptname, bitval) \
static const FieldDesc VMField_##cls##_##scriptname = { #cls, #scriptname, (unsigned)myoffsetof(cls, name), (unsigned)sizeof(cls::name), bitval }; \
extern FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMField_##cls##_##scriptname##_HookPtr GCC_FSEG = &VMField_##cls##_##scriptname;
- added some syntactic help to the ZScript parser to allow defining the arrays with native structs on the script side instead of having to define them internally.
2017-03-13 11:48:25 +00:00
#define DEFINE_GLOBAL(name) \
- fixed variable size mismatch and some warnings for 32 bit builds.
2017-03-14 17:48:51 +00:00
static const FieldDesc VMGlobal_##name = { "", #name, (size_t)&name, (unsigned)sizeof(name), 0 }; \
- added a dummy struct named '_' to define global variables. This can only be used internally. This method was chosen because it avoids adding variable declarations to the global namespace which would have required a lot more work while polluting the grammar. This way the global variables can be handled by a small bit of special coding in the struct generator.
2017-03-13 13:42:14 +00:00
extern FieldDesc const *const VMGlobal_##name##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMGlobal_##name##_HookPtr GCC_FSEG = &VMGlobal_##name;
#define DEFINE_GLOBAL_NAMED(iname, name) \
- fixed variable size mismatch and some warnings for 32 bit builds.
2017-03-14 17:48:51 +00:00
static const FieldDesc VMGlobal_##name = { "", #name, (size_t)&iname, (unsigned)sizeof(iname), 0 }; \
- added some syntactic help to the ZScript parser to allow defining the arrays with native structs on the script side instead of having to define them internally.
2017-03-13 11:48:25 +00:00
extern FieldDesc const *const VMGlobal_##name##_HookPtr; \
MSVC_FSEG FieldDesc const *const VMGlobal_##name##_HookPtr GCC_FSEG = &VMGlobal_##name;
- fixed: CALL_ACTION always set 3 args, even for normal methods. Also moved this to a sunfunction because the macro created a lot of code.
2016-11-06 00:34:54 +00:00
class AActor;
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
- stop using the ATAGs for checking pointer types in asserts. This is an incredibly costly way to do a debug check as it infests the entire VM design from top to bottom. These tags are basically useless for anything else but validating object pointers being passed to native functions (i.e. mismatches between definition and declaration) and that simply does not justify a feature that costs execution time in non-debug builds and added memory overhead everywhere. Note that this commit does not remove the tags, it only discontinues their use.
2017-04-10 13:17:39 +00:00
#define ACTION_RETURN_STATE(v) do { FState *state = v; if (numret > 0) { assert(ret != NULL); ret->SetPointer(state); return 1; } return 0; } while(0)
#define ACTION_RETURN_POINTER(v) do { void *state = v; if (numret > 0) { assert(ret != NULL); ret->SetPointer(state); return 1; } return 0; } while(0)
#define ACTION_RETURN_OBJECT(v) do { auto state = v; if (numret > 0) { assert(ret != NULL); ret->SetObject(state); return 1; } return 0; } while(0)
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#define ACTION_RETURN_FLOAT(v) do { double u = v; if (numret > 0) { assert(ret != nullptr); ret->SetFloat(u); return 1; } return 0; } while(0)
- scriptified A_PainShootSkull which was the last remaining bit in g_doom, so this directory is gone now.
2016-11-23 18:47:09 +00:00
#define ACTION_RETURN_VEC2(v) do { DVector2 u = v; if (numret > 0) { assert(ret != nullptr); ret[0].SetVector2(u); return 1; } return 0; } while(0)
- scriptified a_archvile.cpp. - fixed the type checks for the conditional operator.
2016-10-30 17:41:39 +00:00
#define ACTION_RETURN_VEC3(v) do { DVector3 u = v; if (numret > 0) { assert(ret != nullptr); ret[0].SetVector(u); return 1; } return 0; } while(0)
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
#define ACTION_RETURN_INT(v) do { int u = v; if (numret > 0) { assert(ret != NULL); ret->SetInt(u); return 1; } return 0; } while(0)
#define ACTION_RETURN_BOOL(v) ACTION_RETURN_INT(v)
- implemented code generation for stack variables. - fixed code generation for using local variables as array index. This must use a different register for the array element offset because the original register may not be overwritten.
2016-11-17 15:44:41 +00:00
#define ACTION_RETURN_STRING(v) do { FString u = v; if (numret > 0) { assert(ret != NULL); ret->SetString(u); return 1; } return 0; } while(0)
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
// Checks to see what called the current action function
#define ACTION_CALL_FROM_ACTOR() (stateinfo == nullptr || stateinfo->mStateType == STATE_Actor)
#define ACTION_CALL_FROM_PSPRITE() (self->player && stateinfo != nullptr && stateinfo->mStateType == STATE_Psprite)
#define ACTION_CALL_FROM_INVENTORY() (stateinfo != nullptr && stateinfo->mStateType == STATE_StateChain)
- moved the VM types into their own file and only include it where really needed.
2017-04-12 23:12:04 +00:00
// Standard parameters for all action functions
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
// self - Actor this action is to operate on (player if a weapon)
// stateowner - Actor this action really belongs to (may be an item)
// callingstate - State this action was called from
#define PARAM_ACTION_PROLOGUE(type) \
PARAM_PROLOGUE; \
- fixed: 'self' may never be null for native VM functions so it must always abort if null gets passed.
2017-06-05 06:43:12 +00:00
PARAM_OBJECT_NOT_NULL (self, AActor); \
- corrected the export signatures of several script exports to contain the correct classes, so that the fudging in FindFunction could be removed. - fixed PARAM_ACTION_PROLOGUE to assign correct types to the implicit pointers. It gave the actual class to the wrong one, which until now did not matter because all functions were using 'Actor', regardless of actual class association. - fixed the definition of IceChunk and removed some redundant code here. Since A_FreezeDeathChunks already calls SetState, which in turn calls the state's action function, there is no need to call it again explicitly.
2016-12-02 17:52:58 +00:00
PARAM_OBJECT (stateowner, type) \
- scriptified A_FatAttack*. - swapped parameters of two-parameter VelToAngle method, so that internal and script version are in line. - fixed parameter asserts to handle NULL pointers properly.
2016-10-27 15:47:46 +00:00
PARAM_POINTER (stateinfo, FStateParamInfo) \
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
// Number of action paramaters
#define NAP 3
#define PARAM_SELF_PROLOGUE(type) \
PARAM_PROLOGUE; \
- fixed: 'self' may never be null for native VM functions so it must always abort if null gets passed.
2017-06-05 06:43:12 +00:00
PARAM_OBJECT_NOT_NULL(self, type);
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
- most ATAG stuff is gone, except for the static storage space in the VMFunction.
2017-04-10 14:06:18 +00:00
// for structs we cannot do a class validation
- scriptified the pistol to test if struct member functions work. - made APlayerPawn::PlayAttacking(2) virtual script functions so that mods have better control over player animations. Note that these have no native base so they skip the templated interface for managing virtual functions.
2016-11-19 12:56:29 +00:00
#define PARAM_SELF_STRUCT_PROLOGUE(type) \
PARAM_PROLOGUE; \
- fixed: 'self' may never be null for native VM functions so it must always abort if null gets passed.
2017-06-05 06:43:12 +00:00
PARAM_POINTER_NOT_NULL(self, type);
- use the function defaults from the script instead of explicitly setting them again in the code. This is a needless cause of potential errors and since the values are readily available now it's better to use them in the functions. - fixed: ZCCCompiler did not process array access nodes. - fixed: Function argument names were not placed in the destination list by the compiler. - scriptified several trivial functions from p_actionfunctions.cpp.
2016-10-27 13:53:53 +00:00
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
class PFunction;
- fixed damage handling in A_BetaSkullAttack. For unknown reasons this completely bypassed the normal damage function semantics and even multiplied that with a random value. - made some tests about calling script code from native functions. * scriptified A_SkullAttack to have something to test * changed the A_SkullAttack call in A_PainShootSkull. * use a macro to declare the function pointer. Using local static variable init directly results in hideous code for the need of being thread-safe (which, even if the engine was made multithreaded is not needed here.) * Importsnt node here: Apparently passing an actor pointer to the VMValue constructor results in the void * version being called, not the DObject * version.
2016-11-06 10:36:12 +00:00
VMFunction *FindVMFunction(PClass *cls, const char *name);
#define DECLARE_VMFUNC(cls, name) static VMFunction *name; if (name == nullptr) name = FindVMFunction(RUNTIME_CLASS(cls), #name);
- started with cleanup and separation of DECORATE code. * everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation. * a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
2016-10-12 17:22:33 +00:00
- fixed issues with option menu items. - fixed the octal parser in strbin. - remove 'new' token because it gets in the way.
2017-02-12 15:02:55 +00:00
FString FStringFormat(VM_ARGS);
- separated the code generation from the DECORATE parser and cleaned up the interface to the code generator. Most importantly, the VMScriptFunctions are now preallocated when being added to the list of functions to compile and will be filled in later by the code generator. This allowed the removal of some ugly maintenance code.
2016-10-12 22:53:59 +00:00
- moved the VM types into their own file and only include it where really needed.
2017-04-12 23:12:04 +00:00
unsigned GetVirtualIndex(PClass *cls, const char *funcname);
#define IFVIRTUALPTR(self, cls, funcname) \
static unsigned VIndex = ~0u; \
if (VIndex == ~0u) { \
VIndex = GetVirtualIndex(RUNTIME_CLASS(cls), #funcname); \
assert(VIndex != ~0u); \
} \
auto clss = self->GetClass(); \
VMFunction *func = clss->Virtuals.Size() > VIndex? clss->Virtuals[VIndex] : nullptr; \
if (func != nullptr)
#define IFVIRTUAL(cls, funcname) IFVIRTUALPTR(this, cls, funcname)
#define IFVIRTUALPTRNAME(self, cls, funcname) \
static unsigned VIndex = ~0u; \
if (VIndex == ~0u) { \
VIndex = GetVirtualIndex(PClass::FindClass(cls), #funcname); \
assert(VIndex != ~0u); \
} \
auto clss = self->GetClass(); \
VMFunction *func = clss->Virtuals.Size() > VIndex? clss->Virtuals[VIndex] : nullptr; \
if (func != nullptr)
Normalize line endings
2016-03-01 15:47:10 +00:00
#endif
Reference in a new issue Copy permalink